@jakkrichm/create-nexus-devflow 2.0.0 → 2.0.2

package/template/.claude/skills/specialist-agent-routing/SKILL.md

@@ -1,107 +1,107 @@
----
-name: specialist-agent-routing
-description: Route a request to the right specialist agent and preserve the DevFlow 2.0 stage-first contract. Use when the user wants direct specialist judgment on a file, folder, workspace, or concern.
----
-
-# Specialist Agent Routing
-
-## Overview
-
-This skill is the shared behavior layer behind `Agent`.
-
-It keeps direct specialist invocation available for advanced use, but makes clear that the owning workflow or mainline stage still controls lifecycle state.
-
-## When to Use
-
-- when the user explicitly wants a specialist persona
-- when a stage needs deeper focused judgment on a bounded target
-- when a file, folder, running ID workspace, or codebase concern maps better to one expert than to a whole workflow
-
-## Core Rule
-
-`Agent` is not a mainline stage and should not replace the owning workflow.
-
-Use it to add focused expertise, then route back to the correct lifecycle step.
-
-## Input Shape
-
-```text
-Agent {AGENT_NAME} {TARGET}
-```
-
-Examples:
-
-```text
-Agent requirements-engineer devflow/runs/007/spec.md
-Agent codebase-explorer src/services/
-Agent code-reviewer devflow/runs/007
-```
-
-## Process
-
-### 1. Load The Specialist
-
-Read:
-
-- `.agent/agents/{AGENT_NAME}.md`
-- the minimum necessary target context
-- relevant stage artifacts when the target belongs to a running ID
-
-### 2. Preserve Stage-First Contract
-
-- prefer `discover.md`, `define.md`, `spec.md`, `plan.md`, `implement.md`, `verify.md`, `release.md`, and `report.md`
-- do not silently mutate legacy JSON contracts
-- if the agent is only reviewing, return recommendations or save a report instead of pretending to execute the whole lifecycle
-
-### 3. Apply Specialist Discipline
-
-Examples:
-
-- `code-reviewer` -> correctness, risk, maintainability, findings-first review
-- `prp-core-debugger` -> RCA and validation evidence
-- `requirements-engineer` -> scope, ambiguity, acceptance criteria
-- `codebase-explorer` -> architecture, pattern, and data flow investigation
-
-### 4. Save A Reusable Report When Needed
-
-For substantial outputs, save to:
-
-```text
-devflow/reports/{date}-{agent-name}-{timestamp}.md
-```
-
-Use `.agent/resources/schemas/agent_report.template.md` unless the invoking workflow provides a more specific template.
-
-### 5. Route Back
-
-Return to the owning step:
-
-- `/10-Define`
-- `/20-Spec`
-- `/30-Plan`
-- `/40-Implement`
-- `/50-Verify`
-- `/70-Release`
-- `/60-Report`
-- `PR-Review`
-- `Research`
-
-## Skills That Should Not Be Invoked As Agents
-
-These former narrow agents now belong in the skill layer:
-
-- `code-simplification`
-- `type-design`
-- `silent-failure-audit`
-
-Invoke them through the responsible workflow or agent instead.
-
-## Output
-
-Return:
-
-- agent used
-- target reviewed
-- concise findings or changes
-- report path when a substantial report was written
-- exact recommended next route
+---
+name: specialist-agent-routing
+description: Route a request to the right specialist agent and preserve the DevFlow 2.0 stage-first contract. Use when the user wants direct specialist judgment on a file, folder, workspace, or concern.
+---
+
+# Specialist Agent Routing
+
+## Overview
+
+This skill is the shared behavior layer behind `Agent`.
+
+It keeps direct specialist invocation available for advanced use, but makes clear that the owning workflow or mainline stage still controls lifecycle state.
+
+## When to Use
+
+- when the user explicitly wants a specialist persona
+- when a stage needs deeper focused judgment on a bounded target
+- when a file, folder, running ID workspace, or codebase concern maps better to one expert than to a whole workflow
+
+## Core Rule
+
+`Agent` is not a mainline stage and should not replace the owning workflow.
+
+Use it to add focused expertise, then route back to the correct lifecycle step.
+
+## Input Shape
+
+```text
+Agent {AGENT_NAME} {TARGET}
+```
+
+Examples:
+
+```text
+Agent requirements-engineer devflow/runs/007/spec.md
+Agent codebase-explorer src/services/
+Agent code-reviewer devflow/runs/007
+```
+
+## Process
+
+### 1. Load The Specialist
+
+Read:
+
+- `.agent/agents/{AGENT_NAME}.md`
+- the minimum necessary target context
+- relevant stage artifacts when the target belongs to a running ID
+
+### 2. Preserve Stage-First Contract
+
+- prefer `discover.md`, `define.md`, `spec.md`, `plan.md`, `implement.md`, `verify.md`, `release.md`, and `report.md`
+- do not silently mutate legacy JSON contracts
+- if the agent is only reviewing, return recommendations or save a report instead of pretending to execute the whole lifecycle
+
+### 3. Apply Specialist Discipline
+
+Examples:
+
+- `code-reviewer` -> correctness, risk, maintainability, findings-first review
+- `prp-core-debugger` -> RCA and validation evidence
+- `requirements-engineer` -> scope, ambiguity, acceptance criteria
+- `codebase-explorer` -> architecture, pattern, and data flow investigation
+
+### 4. Save A Reusable Report When Needed
+
+For substantial outputs, save to:
+
+```text
+devflow/reports/{date}-{agent-name}-{timestamp}.md
+```
+
+Use `.agent/resources/schemas/agent_report.template.md` unless the invoking workflow provides a more specific template.
+
+### 5. Route Back
+
+Return to the owning step:
+
+- `/10-Define`
+- `/20-Spec`
+- `/30-Plan`
+- `/40-Implement`
+- `/50-Verify`
+- `/70-Release`
+- `/60-Report`
+- `PR-Review`
+- `Research`
+
+## Skills That Should Not Be Invoked As Agents
+
+These former narrow agents now belong in the skill layer:
+
+- `code-simplification`
+- `type-design`
+- `silent-failure-audit`
+
+Invoke them through the responsible workflow or agent instead.
+
+## Output
+
+Return:
+
+- agent used
+- target reviewed
+- concise findings or changes
+- report path when a substantial report was written
+- exact recommended next route